GitLab Transcend - Intelligent orchestration
Mini toolsFunAudio ASR
By Alibaba
FunAudio-ASR is a large-scale ASR system and model family focused on practical deployment rather than benchmark-only performance. Public model pages describe it as trained on tens of millions of hours of real speech data, with strong contextual understanding and industry adaptability, especially for specialized domains such as education and finance. Current public Fun-ASR model pages also describe low-latency real-time transcription and multilingual coverage up to 31 languages for the newer multilingual Nano variant, while the technical report frames the broader system as an LLM-based ASR stack optimized for streaming, noise robustness, code-switching, hotword customization, and reduced hallucination in real-world use.
New
Multimodal
Gen 3
Overview
FunAudio-ASR, also called Fun-ASR, is Tongyi Lab’s end-to-end speech recognition model family for production ASR. It is trained on very large real-speech data, supports low-latency real-time transcription, and is designed for strong contextual understanding, domain terminology handling, and multilingual recognition.
About Alibaba
Chinese e-commerce and cloud leader behind Taobao, Tmall, and Alipay.
View Company ProfileTools using FunAudio ASR
No tools found for this model yet.
Last updated: April 21, 2026
